AECD-3260 PILE TRANSFER FUNCTIONS. _Mai. By " Joseph P. Franz. July 18, Westinghouse Atomic Power Division UWIXMT!OISTA!

Transcription

1 UNITED STATES ATOMIC ENERGY COMMISSION AECD-3260 PILE TRANSFER FUNCTIONS _Mai By " Joseph P. Franz C-fl UWIXMT!OISTA!TEMEN A~pwrovea tot purzhc twooaso A July 18, 1949 *'IGY CD, Westinghouse Atomic Power Division UTES * L Technical Information Service, Oak Ridge, Tennessee!"My

2 DISCLAIMER NOTICE THIS DOCUMENT IS BEST QUALITY AVAILABLE. THE COPY FURNISHED TO DTIC CONTAINED A SIGNIFICANT NUMBER OF PAGES WHICH DO NOT REPRODUCE LEGIBLY.

3 PHYSICS Issuance of this document does not constitute authority for declassification of classified copies of the same or similar content and title and by the same author(s). Reproduced direct from copy as submitted to this office. XRIN EDUSA Ailab1 from th Office o Tec ical Servl es Depart ent f Commerc Washin o 25, D. C. This report is based on Memo APIC-4 SM-I. Date Declassified: October 26, AEC,Oak Ridge,Tenn., W7830

4 AECD PILE TRANSFER FUNCTIONS The purpose of this memo is to present a detailed derivation of the transfer function of a critical pile and to relate this steady-state description and the transient response to a step function change of reactivity. Reference to G.E. report JIO-1 - Revised is implied throughout this discussion. The transfer function of a critical pile can be determined from the differential equations of the pile. The reactivity of the pile is assumed to be controlled by the delayed neutrons so that keff is always less than 1 + B For these assumptions elementary pile theory gives - where: fi Bi/B / (D - 1)n A J i/ (1) B dt ii d Ci/B _fi na ClB (2) dt T n = neutron flux I - mean life of a neutron Bi= % of group-i fission neutrons B Ci Ai = % of neutrons emitted which are delayed concentration of fission fragments which emit group-i delayed neutrons decay constant of group-i neutrons t = time (seconds) Sk D = reactivity in dollars B Skeff keff - 1 keff - ratio of fission neutrons produced to those lost for an infinite pile The discussion thus far is somewhat misleading. We have been considering a pile on the basis of elementary pile theory, when actually we are going to use an electronic pile simulator whose correspondence to a pile depends on the accuracy of the assumptions of elementary pile theory. To avoid considering these assumptions we will be more exact and say that we desire to determine the transfer function of the electronic pile simulator. The simulator -3-

5 - 4 - AECD-3260 obeys the same differential equations. To determine the transfer function consider the simulator as a four-terminal network. The input function is the reactivity, D = 8 (t) and the output, is the neutron flux. Por a sinusoidal function of time input the transfer function KG(s) is the ratio of the Laplace transforms of the output and input. It consists of K = a constant, or frequency invariant portion, and G(s) = the frequency variant portion, or the portion dependent on time derivatives. In general, we define f(s) as the Laplace transform of f(t) by this integral: f(s) 0 1e-st f(t) dt n(t) e) rg(s) n'(s) 6(s) KG(s) = n'(s) Let the reactivity nave small sinusoidal oscillations of amplitude D = 6 (t) = E e jwt (3) Since the pile is critical, the neutron flux will vary sinusoidally [nj] about a steady level, no. The same is true for the concentrations of fission fragments. Thus: n(t) =no ; n'ct) (4) Ci~t) =Cio A Ci, (t) (5) By definition the transfer function relates sinusoidal variations, so that KG(s) =AL(e (6) To obtain this result we eliminate Ci from equations (1) and (2), take the Laplace transform and substitute to obtain an expression for n'. From (2) ii = di / (7) B J dt

6 - AECD-326o Substituting in (1) and using (3): dn _( 1)n fi n- jd(ci/b3 (8) B dt ii dt By (4) i(5): ýln = hos ý L' n' 8 Sdn'= - n f fn in _ d(i/ (9) B dt i i dt but n = fin It is convenient to neglect n'(t) 8(t). For power levels greater than one quarter power nos (t)»> n'(t) S (t), so that this assumption will not introduce appreciable error. Taking the Laplace transform of (9) gives s n'(s) = nos(s) -i F-ci(s) (10) Rewriting eq. (2) in terms of n'(t) and Cil(t) we have: d CfIB fi f i oi l dt 7n o n. - B 1) But the steady-state condition dci/b dt 0 shows that fi no Biio -7-= B (12) taking the Laplace transform and then solving for C 1 : 8 Co'(s) = n'cs) (f - i Ci(s) B 7 B C = f n'(s) (13) B '(s, Xi) Substituting (13) in (10) L s /, n'(s) nos(s) n'(s) - Hns Bn ^ ( 15) S s B i tz7x (14)

7 -6 - AECD-3260 This can be rewritten: nob F (s (s)(s/2)... (s 0\0)-. (16) 1 (s ao A als A a 2 s L aisi L... - o n, l(js/ -, s(s A sl)(s A S2)..Cs A SO).. (18) This is the general form of the transfer function. All the Al's and si's are positive. That si is positive follows from the fact that all the ails being positive excludes positive roots. This just says that the pile transfer function can be handled by the standard techniques of servo theory. (This was implicit in the original assumptions). This general result may now be applied to a simulator having five groups of delayed neutrons. &if-life Decay constant Fraction of Fission Neutrons tl/2 A i P sec- 1 Bi tl/ )Il 14 BI = X2 = 1.61 B 2 = X3 = B 3 = A4 = B 4 = X5 = B 5 = Assume 5 x 10-5 sec Eq. (17) is now n'(s) nob (s X 1 ))(s A X 2 )(s AX- 3 )(s AX 4 )(s AX5) (19) (s) s s5 A a 4 s4 A a 3 s3, a 2 s2 7 als A ao The ai's can be written in terms of Ai, Bi and/by comparing them with the corresponding coefficients of the denominator expressed in eq. (16).

8 -7 - AECD-3260 The results are shown in Appendix I. Now the roots of this fifth degree equation must be determined. D(s) 9 s A 2270 s a ý73.2=0 (20) (s 7 ( / 2? 53)(8 L 54)(s 7 95) (21) The results are: si = 148 s2 = 13.3 s3 = 1.45 s4 = 0.32 s5 = 0.08 These results can be used to plot the Bode diagram of Figure 1 by replacing s by jw. The Bode diagram (Figure 1) shows the asymptotes of the steady-state amplitude gain of the pile simulator for a sinusoidal input. Replacing s by jw is an obvious operation, since s may be considered equivalent to the time derivative. Taking the time derivative of a sinusoid, exp jwt, is just multiplication by jw. The values of si and N are the break points of the asymptotes of the function whose analytical expression is: LmIKG(jv)l =Lmno A Lm LIM jw A 1 48j (22) A Lm liv A 141 L njiw A , Lmjw A Lm I jw A SIMjw, A Lm jw, SIJw A Lm jw A 0.o81 A Li-~ 0011 ji Transient Response The response of a steady-state pile to a step function change of reactivity, S, can be expressed as a sum of exponential terms. Usually SAie Pi t (23)

9 -8 - AECD-3260 we are interested in six groups of delayed neutrons giving seven terms in the expansion. For each amplitude, 8, there are seven values of p. Each p has an A associated with it. Figure 2 shows the calculated values of Pi and Ai for values of reactivity,, from - $2.00 to L $1.00 (-1.5% to A 0.75%). These curves also are plotted for various values of the mean life of a neutron, -. These step-function response graphs can be related to the Bode diagram of steady-state sinusoidal response. The transfer function of the simulator KG(s) depends only on the simulator's physical characteristicsý. On the other hand the constants obtained from the step-function response graphs depend on the amplitude of the input. If we apply a unit impulse function (unit step-function minus another unit stepfunction) we can use the step-response graphs to determine the output. The values of Ai and Pi are those corresponding to 6 = 0, and in a sense do not depend on reactivity, but only on the characteristics of the simulator. Now by previous definitions - n(s) = S(s)KG(s) (24) If S(t) is an impulse function S(s) = 1 and n(s)i (t)=o = KG(s) (25) The step-function response n(t).no A. Pit (26) has the Laplace transform n(s) = noý 6 Ai pi (27) i11 subject to certain mathematical restrictions (the physical equivalent of not allowing the pile to blow up). For the case of 6 groups of delayed neutrons

10 9 - AECD-326o the transfer function KGs - n 0 (s Al (a ) 2 ) (s /A 3 )s O (s ýx5)s A NL) 6) (8 ( -j s(s A 8 1 )(s A s 2 )(s s3)(5 A s4)(s A, 5 )(s A s 6 ) Equation (25) states that this equals n (s) (n) o F: 6 A "(" =1 s -Pi 116s 6,Lb55 5, b 4 js 4 b 3 s 3 A b 2 s 2 bis ý bn (29) = no (s-pl)(s-p 2 )(s-p 3 )(s-p,)(s-p 5 )(s-p 6 )(s-p 7 ) There is a correspondence between the si s and the pi Is. p 1 = 0 P2 = -O.u15 86 = P3 = s5 = 0.08 P 4 = s4 = 0.32 p5 n S3 = 1.45 P6 = = 13.3 P7 = - si = 148 The value of S6 was obtained from the Bode diagram of G.E. report JIO-l. The values of p, to P6 and 86 to 82 do not change appreciably for different values of neutron life!, so that the above comparison is valid even though the pi's were computed forj= 10-5 and the si's for/= 5 x The value of s1 (and p7) varies considerably for different values of ], but for a given A, s 1 is the negative of p 7. This correspondence between steady-state and transient response gives a convenient check and can be used to correlate results.

11 -10 - AECD apenc&-x.L a + +\ 1+ +x~ + ±ý(x+x 3y + 3 A-\V 4-( +)\,5)\+ + X~? + +X, XtŽ+\ +,? x~( I 3 A 2 ± + 5 +~ TJ

12 - ii -ASECD w z w OU) ckco 0 'd pip '4' 10 ca 0 z 0.J0I..LAft 0n - Hc (nd >-i-( UJ W ~D 00 u) ~

13 -1-AECD-326oG z bu U) 0 j f r 0ý -: w to ( ow tw0 ILL 0 (n a- En U)~T ~ 00 Z 4 In 0 ld > w- 0 7 ir -J w 2- -'m -I w U) a-P<+ 00 W CL ac 0 < X Z0 z +-T Z - oa ' uh :u - al i- - 0,. 0 I~ ~ U -0~- IN0 a- > 3: I)Wf ej I o 0 X4 0 o it fi 1 1 0ý -- < I: I Ir 0 C3 I.- <1 i 0r '0i 0f Wc u- CA 0 I* C" - - U rowl U: i 0 ~ OF 0OIJU